Predictability of slip events along a laboratory fault

A "long" sequence of stick‐slip events generated along a laboratory fault, which consists of eight spring‐connected masses that are elastically driven to slide on a frictional surface, has been examined to check whether the "large" events are predictable. The large events are found to recur at intervals of very different durations, although the elastic and frictional properties along the fault are quite uniform. The recurrence intervals are, however, approximately proportional to the displacements of the preceding events, but not of the following events. Thus the occurrence time of a large event is approximately predictable, whereas its displacement is not. The size (logarithmic energy release) distribution of all the events of different sizes can be fitted by a maximum‐entropy model, which has an upper bound in size corresponding to the complete release of the maximum strain energy stored along the fault.